LLM output compare

Client

Line-by-line comparison for two pasted answers. For generic text, the text diff tool is equivalent—this page frames the workflow around assistant drafts.

When to use it

Use output compare when you have two versions of the same task (different runs, prompts, or models) and want a quick visual on what moved—without sending either version to a server.

Two outputs

?

Line-based diff only (not JSON structure). For structured JSON use JSON compare; for arbitrary plain text use text diff.

Unified view

Summary
- Use UTF-8 for all new files.
- Prefer small modules over giant files.
- Prefer small modules; avoid circular imports.

Next steps
Run the formatter before opening a PR.Run the formatter and tests before opening a PR.

Common use cases

  • Compare two answers after changing only the system prompt or temperature.
  • Sanity-check regressions when a model or settings update changes wording.
  • Copy a unified diff into a ticket without pasting full secrets—redact first.

Common mistakes to avoid

  • Expecting JSON-aware alignment

    This is plain line diff. Use JSON compare when the payloads are structured and key order should not drown the signal.

  • Treating whitespace-only edits as “no change”

    Trailing spaces and blank lines matter in line diff mode—normalize drafts if you only care about semantic text.

FAQ

Is my text uploaded?

No. Diffing runs in your browser with the diff library.

How do agents prefill both sides?

Use the same query keys as plain text diff: left / leftqb and right / rightqb (Base64 when needed).

Common search terms

Phrases people search for that match this tool. See the full long-tail keyword index.

  • compare two llm outputs
  • diff assistant answers offline

More tools

Related utilities you can open in another tab—mostly client-side.